A Guide to CPI Fraud and How To Stop Fraudulent Conversions.

Cost Per Install Fraud (CPI)
ShareShare on FacebookTweet about this on TwitterShare on LinkedIn

Your goal is high qaulity traffic? – How can you reach that goal and stop CPI Fraud? 


Mobile Affiliate Networks who work on a CPI (Cost Per Install) basis provide offers that pay for every app install made. App Developers, Affiliates and media buyers use this opportunity to earn money by promoting these offers and products. 
With an average CPI payout, ranging from $1 – $10, this business can be a lucrative deal (Source: Offervault.com, March 2017).


The nature of fraudsters


Whether we like it or not, a profitable business is always going to attract fraudsters who are trying to make quick money.

Regarding the high advertising budgets provided by companies, fraud is a growing problem within the industry and it is one of the most important issues a marketer has to address when analyzing the quality of their ad campaigns.

It is even more important than viewability, conversion rate or any other metric as it can be a major threat to your business if not addressed properly. According to a brand new report by CNBC, advertising fraud will cost companies up to $16.4 billion in 2017, which is much higher than most experts projected.

If you don’t want to loose your hard earned money to fraudsters, you need to take action.

 

Here are a few tips that you can implement in order to protect your business from fraudulent activities:

Smart Server Farms / Install Bots


This can be described as a software that is imitating human behavior. It automates an install of an app, or even re-installs it if the app has already been installed.

How to identify Fraud by Bots/Emulators:
The best solution to tackle this problem is to link an install with a unique identifier of a

device, also known as ‘Device ID’. Fortunately, there is only one.
If you deal with an experienced fraudster, it is possible to create different IDs from the same Device, which means that you may notice a large number of installs from the same IP as well as irrational timestamps.


Click Stuffing / Click Spam

Click Spam, also known as Click Stuffing describes the practice of attributing an app install to a fake source , which is mostly done via a fraudulent app that has been developed by a fraudster (e.g. Flashlight App). We have talked about this topic in more detail here.
Remember that Click Spam is one of the fastest growing fraud types in the industry across all verticals. Don’t underestimate sophisticated strategies of fraudsters.


Analyze Post-Install Behavior Events

Looking at the post-install behavior will benefit your fraud detection efforts immensely. The higher you go up in the conversion funnel, it becomes more difficult to fake human behaviour. By looking at the post install data you will be able to easier identify fraudulent patterns.

If you are for example seeing specific users coming from the same source with no post-install in-app activities such as clicks, or other activities inside the app this source can be flagged as fraud.
Finally you want to get real users who like and use the app. If you want to maximize your CPI Payouts, you also need to pay attention what users do after the download. Are they actually using the app? What is the behaviour once they have the app on their device?

How can machine learning algorithms help you?

Fraud Detection is not easy and can be a tedious task. Especially when you are facing a daily volume of multiple thousands of installs.

You most likely need to hire a team to take care of this or use a fraud detection software.

Fortunately we live in a time of machine learning algorithms.
If you are not planning to hire more employees to take care of your fraud problems and if you want to find out more about how an artificial intelligence and machine learning can help you solve this, contact us.

At 24metrics, we are curious to see what type of fraud problems you have seen and what difficulties you encountered in the past.
We love to hear from you, contact us here.